You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

622 lines
25KB

  1. /*
  2. * Copyright (c) 2010, Google, Inc.
  3. *
  4. * This file is part of FFmpeg.
  5. *
  6. * FFmpeg is free software; you can redistribute it and/or
  7. * modify it under the terms of the GNU Lesser General Public
  8. * License as published by the Free Software Foundation; either
  9. * version 2.1 of the License, or (at your option) any later version.
  10. *
  11. * FFmpeg is distributed in the hope that it will be useful,
  12. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  14. * Lesser General Public License for more details.
  15. *
  16. * You should have received a copy of the GNU Lesser General Public
  17. * License along with FFmpeg; if not, write to the Free Software
  18. * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  19. */
  20. /**
  21. * @file
  22. * VP8 encoder support via libvpx
  23. */
  24. #define VPX_DISABLE_CTRL_TYPECHECKS 1
  25. #define VPX_CODEC_DISABLE_COMPAT 1
  26. #include <vpx/vpx_encoder.h>
  27. #include <vpx/vp8cx.h>
  28. #include "avcodec.h"
  29. #include "internal.h"
  30. #include "libavutil/base64.h"
  31. #include "libavutil/mathematics.h"
  32. #include "libavutil/opt.h"
  33. /**
  34. * Portion of struct vpx_codec_cx_pkt from vpx_encoder.h.
  35. * One encoded frame returned from the library.
  36. */
  37. struct FrameListData {
  38. void *buf; /**< compressed data buffer */
  39. size_t sz; /**< length of compressed data */
  40. int64_t pts; /**< time stamp to show frame
  41. (in timebase units) */
  42. unsigned long duration; /**< duration to show frame
  43. (in timebase units) */
  44. uint32_t flags; /**< flags for this frame */
  45. struct FrameListData *next;
  46. };
  47. typedef struct VP8EncoderContext {
  48. AVClass *class;
  49. struct vpx_codec_ctx encoder;
  50. struct vpx_image rawimg;
  51. struct vpx_fixed_buf twopass_stats;
  52. int deadline; //i.e., RT/GOOD/BEST
  53. struct FrameListData *coded_frame_list;
  54. int cpu_used;
  55. int auto_alt_ref;
  56. int arnr_max_frames;
  57. int arnr_strength;
  58. int arnr_type;
  59. int lag_in_frames;
  60. int error_resilient;
  61. int crf;
  62. } VP8Context;
  63. /** String mappings for enum vp8e_enc_control_id */
  64. static const char *ctlidstr[] = {
  65. [VP8E_UPD_ENTROPY] = "VP8E_UPD_ENTROPY",
  66. [VP8E_UPD_REFERENCE] = "VP8E_UPD_REFERENCE",
  67. [VP8E_USE_REFERENCE] = "VP8E_USE_REFERENCE",
  68. [VP8E_SET_ROI_MAP] = "VP8E_SET_ROI_MAP",
  69. [VP8E_SET_ACTIVEMAP] = "VP8E_SET_ACTIVEMAP",
  70. [VP8E_SET_SCALEMODE] = "VP8E_SET_SCALEMODE",
  71. [VP8E_SET_CPUUSED] = "VP8E_SET_CPUUSED",
  72. [VP8E_SET_ENABLEAUTOALTREF] = "VP8E_SET_ENABLEAUTOALTREF",
  73. [VP8E_SET_NOISE_SENSITIVITY] = "VP8E_SET_NOISE_SENSITIVITY",
  74. [VP8E_SET_SHARPNESS] = "VP8E_SET_SHARPNESS",
  75. [VP8E_SET_STATIC_THRESHOLD] = "VP8E_SET_STATIC_THRESHOLD",
  76. [VP8E_SET_TOKEN_PARTITIONS] = "VP8E_SET_TOKEN_PARTITIONS",
  77. [VP8E_GET_LAST_QUANTIZER] = "VP8E_GET_LAST_QUANTIZER",
  78. [VP8E_SET_ARNR_MAXFRAMES] = "VP8E_SET_ARNR_MAXFRAMES",
  79. [VP8E_SET_ARNR_STRENGTH] = "VP8E_SET_ARNR_STRENGTH",
  80. [VP8E_SET_ARNR_TYPE] = "VP8E_SET_ARNR_TYPE",
  81. [VP8E_SET_CQ_LEVEL] = "VP8E_SET_CQ_LEVEL",
  82. };
  83. static av_cold void log_encoder_error(AVCodecContext *avctx, const char *desc)
  84. {
  85. VP8Context *ctx = avctx->priv_data;
  86. const char *error = vpx_codec_error(&ctx->encoder);
  87. const char *detail = vpx_codec_error_detail(&ctx->encoder);
  88. av_log(avctx, AV_LOG_ERROR, "%s: %s\n", desc, error);
  89. if (detail)
  90. av_log(avctx, AV_LOG_ERROR, " Additional information: %s\n", detail);
  91. }
  92. static av_cold void dump_enc_cfg(AVCodecContext *avctx,
  93. const struct vpx_codec_enc_cfg *cfg)
  94. {
  95. int width = -30;
  96. int level = AV_LOG_DEBUG;
  97. av_log(avctx, level, "vpx_codec_enc_cfg\n");
  98. av_log(avctx, level, "generic settings\n"
  99. " %*s%u\n %*s%u\n %*s%u\n %*s%u\n %*s%u\n"
  100. " %*s{%u/%u}\n %*s%u\n %*s%d\n %*s%u\n",
  101. width, "g_usage:", cfg->g_usage,
  102. width, "g_threads:", cfg->g_threads,
  103. width, "g_profile:", cfg->g_profile,
  104. width, "g_w:", cfg->g_w,
  105. width, "g_h:", cfg->g_h,
  106. width, "g_timebase:", cfg->g_timebase.num, cfg->g_timebase.den,
  107. width, "g_error_resilient:", cfg->g_error_resilient,
  108. width, "g_pass:", cfg->g_pass,
  109. width, "g_lag_in_frames:", cfg->g_lag_in_frames);
  110. av_log(avctx, level, "rate control settings\n"
  111. " %*s%u\n %*s%u\n %*s%u\n %*s%u\n"
  112. " %*s%d\n %*s%p(%zu)\n %*s%u\n",
  113. width, "rc_dropframe_thresh:", cfg->rc_dropframe_thresh,
  114. width, "rc_resize_allowed:", cfg->rc_resize_allowed,
  115. width, "rc_resize_up_thresh:", cfg->rc_resize_up_thresh,
  116. width, "rc_resize_down_thresh:", cfg->rc_resize_down_thresh,
  117. width, "rc_end_usage:", cfg->rc_end_usage,
  118. width, "rc_twopass_stats_in:", cfg->rc_twopass_stats_in.buf, cfg->rc_twopass_stats_in.sz,
  119. width, "rc_target_bitrate:", cfg->rc_target_bitrate);
  120. av_log(avctx, level, "quantizer settings\n"
  121. " %*s%u\n %*s%u\n",
  122. width, "rc_min_quantizer:", cfg->rc_min_quantizer,
  123. width, "rc_max_quantizer:", cfg->rc_max_quantizer);
  124. av_log(avctx, level, "bitrate tolerance\n"
  125. " %*s%u\n %*s%u\n",
  126. width, "rc_undershoot_pct:", cfg->rc_undershoot_pct,
  127. width, "rc_overshoot_pct:", cfg->rc_overshoot_pct);
  128. av_log(avctx, level, "decoder buffer model\n"
  129. " %*s%u\n %*s%u\n %*s%u\n",
  130. width, "rc_buf_sz:", cfg->rc_buf_sz,
  131. width, "rc_buf_initial_sz:", cfg->rc_buf_initial_sz,
  132. width, "rc_buf_optimal_sz:", cfg->rc_buf_optimal_sz);
  133. av_log(avctx, level, "2 pass rate control settings\n"
  134. " %*s%u\n %*s%u\n %*s%u\n",
  135. width, "rc_2pass_vbr_bias_pct:", cfg->rc_2pass_vbr_bias_pct,
  136. width, "rc_2pass_vbr_minsection_pct:", cfg->rc_2pass_vbr_minsection_pct,
  137. width, "rc_2pass_vbr_maxsection_pct:", cfg->rc_2pass_vbr_maxsection_pct);
  138. av_log(avctx, level, "keyframing settings\n"
  139. " %*s%d\n %*s%u\n %*s%u\n",
  140. width, "kf_mode:", cfg->kf_mode,
  141. width, "kf_min_dist:", cfg->kf_min_dist,
  142. width, "kf_max_dist:", cfg->kf_max_dist);
  143. av_log(avctx, level, "\n");
  144. }
  145. static void coded_frame_add(void *list, struct FrameListData *cx_frame)
  146. {
  147. struct FrameListData **p = list;
  148. while (*p != NULL)
  149. p = &(*p)->next;
  150. *p = cx_frame;
  151. cx_frame->next = NULL;
  152. }
  153. static av_cold void free_coded_frame(struct FrameListData *cx_frame)
  154. {
  155. av_freep(&cx_frame->buf);
  156. av_freep(&cx_frame);
  157. }
  158. static av_cold void free_frame_list(struct FrameListData *list)
  159. {
  160. struct FrameListData *p = list;
  161. while (p) {
  162. list = list->next;
  163. free_coded_frame(p);
  164. p = list;
  165. }
  166. }
  167. static av_cold int codecctl_int(AVCodecContext *avctx,
  168. enum vp8e_enc_control_id id, int val)
  169. {
  170. VP8Context *ctx = avctx->priv_data;
  171. char buf[80];
  172. int width = -30;
  173. int res;
  174. snprintf(buf, sizeof(buf), "%s:", ctlidstr[id]);
  175. av_log(avctx, AV_LOG_DEBUG, " %*s%d\n", width, buf, val);
  176. res = vpx_codec_control(&ctx->encoder, id, val);
  177. if (res != VPX_CODEC_OK) {
  178. snprintf(buf, sizeof(buf), "Failed to set %s codec control",
  179. ctlidstr[id]);
  180. log_encoder_error(avctx, buf);
  181. }
  182. return res == VPX_CODEC_OK ? 0 : AVERROR(EINVAL);
  183. }
  184. static av_cold int vp8_free(AVCodecContext *avctx)
  185. {
  186. VP8Context *ctx = avctx->priv_data;
  187. vpx_codec_destroy(&ctx->encoder);
  188. av_freep(&ctx->twopass_stats.buf);
  189. av_freep(&avctx->coded_frame);
  190. av_freep(&avctx->stats_out);
  191. free_frame_list(ctx->coded_frame_list);
  192. return 0;
  193. }
  194. static av_cold int vp8_init(AVCodecContext *avctx)
  195. {
  196. VP8Context *ctx = avctx->priv_data;
  197. const struct vpx_codec_iface *iface = &vpx_codec_vp8_cx_algo;
  198. struct vpx_codec_enc_cfg enccfg;
  199. int res;
  200. av_log(avctx, AV_LOG_INFO, "%s\n", vpx_codec_version_str());
  201. av_log(avctx, AV_LOG_VERBOSE, "%s\n", vpx_codec_build_config());
  202. if ((res = vpx_codec_enc_config_default(iface, &enccfg, 0)) != VPX_CODEC_OK) {
  203. av_log(avctx, AV_LOG_ERROR, "Failed to get config: %s\n",
  204. vpx_codec_err_to_string(res));
  205. return AVERROR(EINVAL);
  206. }
  207. dump_enc_cfg(avctx, &enccfg);
  208. enccfg.g_w = avctx->width;
  209. enccfg.g_h = avctx->height;
  210. enccfg.g_timebase.num = avctx->time_base.num;
  211. enccfg.g_timebase.den = avctx->time_base.den;
  212. enccfg.g_threads = avctx->thread_count;
  213. #if FF_API_X264_GLOBAL_OPTS
  214. enccfg.g_lag_in_frames= FFMIN(avctx->rc_lookahead, 25); //0-25, avoids init failure
  215. if (ctx->lag_in_frames >= 0)
  216. enccfg.g_lag_in_frames = ctx->lag_in_frames;
  217. #else
  218. enccfg.g_lag_in_frames= ctx->lag_in_frames;
  219. #endif
  220. if (avctx->flags & CODEC_FLAG_PASS1)
  221. enccfg.g_pass = VPX_RC_FIRST_PASS;
  222. else if (avctx->flags & CODEC_FLAG_PASS2)
  223. enccfg.g_pass = VPX_RC_LAST_PASS;
  224. else
  225. enccfg.g_pass = VPX_RC_ONE_PASS;
  226. if (avctx->rc_min_rate == avctx->rc_max_rate &&
  227. avctx->rc_min_rate == avctx->bit_rate)
  228. enccfg.rc_end_usage = VPX_CBR;
  229. #if FF_API_X264_GLOBAL_OPTS
  230. else if (avctx->crf || ctx->crf > 0)
  231. #else
  232. else if (ctx->crf)
  233. #endif
  234. enccfg.rc_end_usage = VPX_CQ;
  235. enccfg.rc_target_bitrate = av_rescale_rnd(avctx->bit_rate, 1, 1000,
  236. AV_ROUND_NEAR_INF);
  237. if (avctx->qmin > 0)
  238. enccfg.rc_min_quantizer = avctx->qmin;
  239. if (avctx->qmax > 0)
  240. enccfg.rc_max_quantizer = avctx->qmax;
  241. enccfg.rc_dropframe_thresh = avctx->frame_skip_threshold;
  242. //0-100 (0 => CBR, 100 => VBR)
  243. enccfg.rc_2pass_vbr_bias_pct = round(avctx->qcompress * 100);
  244. enccfg.rc_2pass_vbr_minsection_pct =
  245. avctx->rc_min_rate * 100LL / avctx->bit_rate;
  246. if (avctx->rc_max_rate)
  247. enccfg.rc_2pass_vbr_maxsection_pct =
  248. avctx->rc_max_rate * 100LL / avctx->bit_rate;
  249. if (avctx->rc_buffer_size)
  250. enccfg.rc_buf_sz =
  251. avctx->rc_buffer_size * 1000LL / avctx->bit_rate;
  252. if (avctx->rc_initial_buffer_occupancy)
  253. enccfg.rc_buf_initial_sz =
  254. avctx->rc_initial_buffer_occupancy * 1000LL / avctx->bit_rate;
  255. enccfg.rc_buf_optimal_sz = enccfg.rc_buf_sz * 5 / 6;
  256. enccfg.rc_undershoot_pct = round(avctx->rc_buffer_aggressivity * 100);
  257. //_enc_init() will balk if kf_min_dist differs from max w/VPX_KF_AUTO
  258. if (avctx->keyint_min >= 0 && avctx->keyint_min == avctx->gop_size)
  259. enccfg.kf_min_dist = avctx->keyint_min;
  260. if (avctx->gop_size >= 0)
  261. enccfg.kf_max_dist = avctx->gop_size;
  262. if (enccfg.g_pass == VPX_RC_FIRST_PASS)
  263. enccfg.g_lag_in_frames = 0;
  264. else if (enccfg.g_pass == VPX_RC_LAST_PASS) {
  265. int decode_size;
  266. if (!avctx->stats_in) {
  267. av_log(avctx, AV_LOG_ERROR, "No stats file for second pass\n");
  268. return AVERROR_INVALIDDATA;
  269. }
  270. ctx->twopass_stats.sz = strlen(avctx->stats_in) * 3 / 4;
  271. ctx->twopass_stats.buf = av_malloc(ctx->twopass_stats.sz);
  272. if (!ctx->twopass_stats.buf) {
  273. av_log(avctx, AV_LOG_ERROR,
  274. "Stat buffer alloc (%zu bytes) failed\n",
  275. ctx->twopass_stats.sz);
  276. return AVERROR(ENOMEM);
  277. }
  278. decode_size = av_base64_decode(ctx->twopass_stats.buf, avctx->stats_in,
  279. ctx->twopass_stats.sz);
  280. if (decode_size < 0) {
  281. av_log(avctx, AV_LOG_ERROR, "Stat buffer decode failed\n");
  282. return AVERROR_INVALIDDATA;
  283. }
  284. ctx->twopass_stats.sz = decode_size;
  285. enccfg.rc_twopass_stats_in = ctx->twopass_stats;
  286. }
  287. /* 0-3: For non-zero values the encoder increasingly optimizes for reduced
  288. complexity playback on low powered devices at the expense of encode
  289. quality. */
  290. if (avctx->profile != FF_PROFILE_UNKNOWN)
  291. enccfg.g_profile = avctx->profile;
  292. enccfg.g_error_resilient = ctx->error_resilient;
  293. dump_enc_cfg(avctx, &enccfg);
  294. /* Construct Encoder Context */
  295. res = vpx_codec_enc_init(&ctx->encoder, iface, &enccfg, 0);
  296. if (res != VPX_CODEC_OK) {
  297. log_encoder_error(avctx, "Failed to initialize encoder");
  298. return AVERROR(EINVAL);
  299. }
  300. //codec control failures are currently treated only as warnings
  301. av_log(avctx, AV_LOG_DEBUG, "vpx_codec_control\n");
  302. if (ctx->cpu_used != INT_MIN)
  303. codecctl_int(avctx, VP8E_SET_CPUUSED, ctx->cpu_used);
  304. if (ctx->auto_alt_ref >= 0)
  305. codecctl_int(avctx, VP8E_SET_ENABLEAUTOALTREF, ctx->auto_alt_ref);
  306. if (ctx->arnr_max_frames >= 0)
  307. codecctl_int(avctx, VP8E_SET_ARNR_MAXFRAMES, ctx->arnr_max_frames);
  308. if (ctx->arnr_strength >= 0)
  309. codecctl_int(avctx, VP8E_SET_ARNR_STRENGTH, ctx->arnr_strength);
  310. if (ctx->arnr_type >= 0)
  311. codecctl_int(avctx, VP8E_SET_ARNR_TYPE, ctx->arnr_type);
  312. codecctl_int(avctx, VP8E_SET_NOISE_SENSITIVITY, avctx->noise_reduction);
  313. codecctl_int(avctx, VP8E_SET_TOKEN_PARTITIONS, av_log2(avctx->slices));
  314. codecctl_int(avctx, VP8E_SET_STATIC_THRESHOLD, avctx->mb_threshold);
  315. #if FF_API_X264_GLOBAL_OPTS
  316. codecctl_int(avctx, VP8E_SET_CQ_LEVEL, (int)avctx->crf);
  317. if (ctx->crf >= 0)
  318. codecctl_int(avctx, VP8E_SET_CQ_LEVEL, ctx->crf);
  319. #else
  320. codecctl_int(avctx, VP8E_SET_CQ_LEVEL, ctx->crf);
  321. #endif
  322. av_log(avctx, AV_LOG_DEBUG, "Using deadline: %d\n", ctx->deadline);
  323. //provide dummy value to initialize wrapper, values will be updated each _encode()
  324. vpx_img_wrap(&ctx->rawimg, VPX_IMG_FMT_I420, avctx->width, avctx->height, 1,
  325. (unsigned char*)1);
  326. avctx->coded_frame = avcodec_alloc_frame();
  327. if (!avctx->coded_frame) {
  328. av_log(avctx, AV_LOG_ERROR, "Error allocating coded frame\n");
  329. vp8_free(avctx);
  330. return AVERROR(ENOMEM);
  331. }
  332. return 0;
  333. }
  334. static inline void cx_pktcpy(struct FrameListData *dst,
  335. const struct vpx_codec_cx_pkt *src)
  336. {
  337. dst->pts = src->data.frame.pts;
  338. dst->duration = src->data.frame.duration;
  339. dst->flags = src->data.frame.flags;
  340. dst->sz = src->data.frame.sz;
  341. dst->buf = src->data.frame.buf;
  342. }
  343. /**
  344. * Store coded frame information in format suitable for return from encode().
  345. *
  346. * Write buffer information from @a cx_frame to @a buf & @a buf_size.
  347. * Timing/frame details to @a coded_frame.
  348. * @return Frame size written to @a buf on success
  349. * @return AVERROR(EINVAL) on error
  350. */
  351. static int storeframe(AVCodecContext *avctx, struct FrameListData *cx_frame,
  352. uint8_t *buf, int buf_size, AVFrame *coded_frame)
  353. {
  354. if ((int) cx_frame->sz <= buf_size) {
  355. buf_size = cx_frame->sz;
  356. memcpy(buf, cx_frame->buf, buf_size);
  357. coded_frame->pts = cx_frame->pts;
  358. coded_frame->key_frame = !!(cx_frame->flags & VPX_FRAME_IS_KEY);
  359. if (coded_frame->key_frame)
  360. coded_frame->pict_type = AV_PICTURE_TYPE_I;
  361. else
  362. coded_frame->pict_type = AV_PICTURE_TYPE_P;
  363. } else {
  364. av_log(avctx, AV_LOG_ERROR,
  365. "Compressed frame larger than storage provided! (%zu/%d)\n",
  366. cx_frame->sz, buf_size);
  367. return AVERROR(EINVAL);
  368. }
  369. return buf_size;
  370. }
  371. /**
  372. * Queue multiple output frames from the encoder, returning the front-most.
  373. * In cases where vpx_codec_get_cx_data() returns more than 1 frame append
  374. * the frame queue. Return the head frame if available.
  375. * @return Stored frame size
  376. * @return AVERROR(EINVAL) on output size error
  377. * @return AVERROR(ENOMEM) on coded frame queue data allocation error
  378. */
  379. static int queue_frames(AVCodecContext *avctx, uint8_t *buf, int buf_size,
  380. AVFrame *coded_frame)
  381. {
  382. VP8Context *ctx = avctx->priv_data;
  383. const struct vpx_codec_cx_pkt *pkt;
  384. const void *iter = NULL;
  385. int size = 0;
  386. if (ctx->coded_frame_list) {
  387. struct FrameListData *cx_frame = ctx->coded_frame_list;
  388. /* return the leading frame if we've already begun queueing */
  389. size = storeframe(avctx, cx_frame, buf, buf_size, coded_frame);
  390. if (size < 0)
  391. return AVERROR(EINVAL);
  392. ctx->coded_frame_list = cx_frame->next;
  393. free_coded_frame(cx_frame);
  394. }
  395. /* consume all available output from the encoder before returning. buffers
  396. are only good through the next vpx_codec call */
  397. while ((pkt = vpx_codec_get_cx_data(&ctx->encoder, &iter))) {
  398. switch (pkt->kind) {
  399. case VPX_CODEC_CX_FRAME_PKT:
  400. if (!size) {
  401. struct FrameListData cx_frame;
  402. /* avoid storing the frame when the list is empty and we haven't yet
  403. provided a frame for output */
  404. assert(!ctx->coded_frame_list);
  405. cx_pktcpy(&cx_frame, pkt);
  406. size = storeframe(avctx, &cx_frame, buf, buf_size, coded_frame);
  407. if (size < 0)
  408. return AVERROR(EINVAL);
  409. } else {
  410. struct FrameListData *cx_frame =
  411. av_malloc(sizeof(struct FrameListData));
  412. if (!cx_frame) {
  413. av_log(avctx, AV_LOG_ERROR,
  414. "Frame queue element alloc failed\n");
  415. return AVERROR(ENOMEM);
  416. }
  417. cx_pktcpy(cx_frame, pkt);
  418. cx_frame->buf = av_malloc(cx_frame->sz);
  419. if (!cx_frame->buf) {
  420. av_log(avctx, AV_LOG_ERROR,
  421. "Data buffer alloc (%zu bytes) failed\n",
  422. cx_frame->sz);
  423. return AVERROR(ENOMEM);
  424. }
  425. memcpy(cx_frame->buf, pkt->data.frame.buf, pkt->data.frame.sz);
  426. coded_frame_add(&ctx->coded_frame_list, cx_frame);
  427. }
  428. break;
  429. case VPX_CODEC_STATS_PKT: {
  430. struct vpx_fixed_buf *stats = &ctx->twopass_stats;
  431. stats->buf = av_realloc_f(stats->buf, 1,
  432. stats->sz + pkt->data.twopass_stats.sz);
  433. if (!stats->buf) {
  434. av_log(avctx, AV_LOG_ERROR, "Stat buffer realloc failed\n");
  435. return AVERROR(ENOMEM);
  436. }
  437. memcpy((uint8_t*)stats->buf + stats->sz,
  438. pkt->data.twopass_stats.buf, pkt->data.twopass_stats.sz);
  439. stats->sz += pkt->data.twopass_stats.sz;
  440. break;
  441. }
  442. case VPX_CODEC_PSNR_PKT: //FIXME add support for CODEC_FLAG_PSNR
  443. case VPX_CODEC_CUSTOM_PKT:
  444. //ignore unsupported/unrecognized packet types
  445. break;
  446. }
  447. }
  448. return size;
  449. }
  450. static int vp8_encode(AVCodecContext *avctx, uint8_t *buf, int buf_size,
  451. void *data)
  452. {
  453. VP8Context *ctx = avctx->priv_data;
  454. AVFrame *frame = data;
  455. struct vpx_image *rawimg = NULL;
  456. int64_t timestamp = 0;
  457. int res, coded_size;
  458. if (frame) {
  459. rawimg = &ctx->rawimg;
  460. rawimg->planes[VPX_PLANE_Y] = frame->data[0];
  461. rawimg->planes[VPX_PLANE_U] = frame->data[1];
  462. rawimg->planes[VPX_PLANE_V] = frame->data[2];
  463. rawimg->stride[VPX_PLANE_Y] = frame->linesize[0];
  464. rawimg->stride[VPX_PLANE_U] = frame->linesize[1];
  465. rawimg->stride[VPX_PLANE_V] = frame->linesize[2];
  466. timestamp = frame->pts;
  467. }
  468. res = vpx_codec_encode(&ctx->encoder, rawimg, timestamp,
  469. avctx->ticks_per_frame, 0, ctx->deadline);
  470. if (res != VPX_CODEC_OK) {
  471. log_encoder_error(avctx, "Error encoding frame");
  472. return AVERROR_INVALIDDATA;
  473. }
  474. coded_size = queue_frames(avctx, buf, buf_size, avctx->coded_frame);
  475. if (!frame && avctx->flags & CODEC_FLAG_PASS1) {
  476. unsigned int b64_size = AV_BASE64_SIZE(ctx->twopass_stats.sz);
  477. avctx->stats_out = av_malloc(b64_size);
  478. if (!avctx->stats_out) {
  479. av_log(avctx, AV_LOG_ERROR, "Stat buffer alloc (%d bytes) failed\n",
  480. b64_size);
  481. return AVERROR(ENOMEM);
  482. }
  483. av_base64_encode(avctx->stats_out, b64_size, ctx->twopass_stats.buf,
  484. ctx->twopass_stats.sz);
  485. }
  486. return coded_size;
  487. }
  488. #define OFFSET(x) offsetof(VP8Context, x)
  489. #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
  490. static const AVOption options[]={
  491. { "cpu-used", "Quality/Speed ratio modifier", OFFSET(cpu_used), FF_OPT_TYPE_INT, {INT_MIN}, INT_MIN, INT_MAX, VE},
  492. { "auto-alt-ref", "Enable use of alternate reference "
  493. "frames (2-pass only)", OFFSET(auto_alt_ref), FF_OPT_TYPE_INT, {-1}, -1, 1, VE},
  494. { "lag-in-frames", "Number of frames to look ahead for "
  495. "alternate reference frame selection", OFFSET(lag_in_frames), FF_OPT_TYPE_INT, {-1}, -1, INT_MAX, VE},
  496. { "arnr-maxframes", "altref noise reduction max frame count", OFFSET(arnr_max_frames), FF_OPT_TYPE_INT, {-1}, -1, INT_MAX, VE},
  497. { "arnr-strength", "altref noise reduction filter strength", OFFSET(arnr_strength), FF_OPT_TYPE_INT, {-1}, -1, INT_MAX, VE},
  498. { "arnr-type", "altref noise reduction filter type", OFFSET(arnr_type), FF_OPT_TYPE_INT, {-1}, -1, INT_MAX, VE, "arnr_type"},
  499. { "backward", NULL, 0, FF_OPT_TYPE_CONST, {1}, 0, 0, VE, "arnr_type" },
  500. { "forward", NULL, 0, FF_OPT_TYPE_CONST, {2}, 0, 0, VE, "arnr_type" },
  501. { "centered", NULL, 0, FF_OPT_TYPE_CONST, {3}, 0, 0, VE, "arnr_type" },
  502. { "deadline", "Time to spend encoding, in microseconds.", OFFSET(deadline), FF_OPT_TYPE_INT, {VPX_DL_GOOD_QUALITY}, INT_MIN, INT_MAX, VE, "quality"},
  503. { "best", NULL, 0, FF_OPT_TYPE_CONST, {VPX_DL_BEST_QUALITY}, 0, 0, VE, "quality"},
  504. { "good", NULL, 0, FF_OPT_TYPE_CONST, {VPX_DL_GOOD_QUALITY}, 0, 0, VE, "quality"},
  505. { "realtime", NULL, 0, FF_OPT_TYPE_CONST, {VPX_DL_REALTIME}, 0, 0, VE, "quality"},
  506. { "error-resilient", "Error resilience configuration", OFFSET(error_resilient), FF_OPT_TYPE_FLAGS, {0}, INT_MIN, INT_MAX, VE, "er"},
  507. #ifdef VPX_ERROR_RESILIENT_DEFAULT
  508. { "default", "Improve resiliency against losses of whole frames", 0, FF_OPT_TYPE_CONST, {VPX_ERROR_RESILIENT_DEFAULT}, 0, 0, VE, "er"},
  509. { "partitions", "The frame partitions are independently decodable "
  510. "by the bool decoder, meaning that partitions can be decoded even "
  511. "though earlier partitions have been lost. Note that intra predicition"
  512. " is still done over the partition boundary.", 0, FF_OPT_TYPE_CONST, {VPX_ERROR_RESILIENT_PARTITIONS}, 0, 0, VE, "er"},
  513. #endif
  514. {"speed", "", offsetof(VP8Context, cpu_used), FF_OPT_TYPE_INT, {.dbl = 3}, -16, 16, VE},
  515. {"quality", "", offsetof(VP8Context, deadline), FF_OPT_TYPE_INT, {.dbl = VPX_DL_GOOD_QUALITY}, INT_MIN, INT_MAX, VE, "quality"},
  516. {"best", NULL, 0, FF_OPT_TYPE_CONST, {.dbl = VPX_DL_BEST_QUALITY}, INT_MIN, INT_MAX, VE, "quality"},
  517. {"good", NULL, 0, FF_OPT_TYPE_CONST, {.dbl = VPX_DL_GOOD_QUALITY}, INT_MIN, INT_MAX, VE, "quality"},
  518. {"realtime", NULL, 0, FF_OPT_TYPE_CONST, {.dbl = VPX_DL_REALTIME}, INT_MIN, INT_MAX, VE, "quality"},
  519. {"arnr_max_frames", "altref noise reduction max frame count", offsetof(VP8Context, arnr_max_frames), FF_OPT_TYPE_INT, {.dbl = 0}, 0, 15, VE},
  520. {"arnr_strength", "altref noise reduction filter strength", offsetof(VP8Context, arnr_strength), FF_OPT_TYPE_INT, {.dbl = 3}, 0, 6, VE},
  521. {"arnr_type", "altref noise reduction filter type", offsetof(VP8Context, arnr_type), FF_OPT_TYPE_INT, {.dbl = 3}, 1, 3, VE},
  522. #if FF_API_X264_GLOBAL_OPTS
  523. {"rc_lookahead", "Number of frames to look ahead for alternate reference frame selection", offsetof(VP8Context, lag_in_frames), FF_OPT_TYPE_INT, {.dbl = -1}, -1, 25, VE},
  524. {"crf", "Select the quality for constant quality mode", offsetof(VP8Context, crf), FF_OPT_TYPE_INT, {.dbl = -1}, -1, 63, VE},
  525. #else
  526. {"rc_lookahead", "Number of frames to look ahead for alternate reference frame selection", offsetof(VP8Context, lag_in_frames), FF_OPT_TYPE_INT, {.dbl = 25}, 0, 25, VE},
  527. {"crf", "Select the quality for constant quality mode", offsetof(VP8Context, crf), FF_OPT_TYPE_INT, {.dbl = 0}, 0, 63, VE},
  528. #endif
  529. {NULL}
  530. };
  531. static const AVClass class = {
  532. .class_name = "libvpx encoder",
  533. .item_name = av_default_item_name,
  534. .option = options,
  535. .version = LIBAVUTIL_VERSION_INT,
  536. };
  537. static const AVCodecDefault defaults[] = {
  538. { "qmin", "-1" },
  539. { "qmax", "-1" },
  540. { "g", "-1" },
  541. { "keyint_min", "-1" },
  542. { NULL },
  543. };
  544. AVCodec ff_libvpx_encoder = {
  545. .name = "libvpx",
  546. .type = AVMEDIA_TYPE_VIDEO,
  547. .id = CODEC_ID_VP8,
  548. .priv_data_size = sizeof(VP8Context),
  549. .init = vp8_init,
  550. .encode = vp8_encode,
  551. .close = vp8_free,
  552. .capabilities = CODEC_CAP_DELAY,
  553. .pix_fmts = (const enum PixelFormat[]){PIX_FMT_YUV420P, PIX_FMT_NONE},
  554. .long_name = NULL_IF_CONFIG_SMALL("libvpx VP8"),
  555. .priv_class = &class,
  556. .defaults = defaults,
  557. };